Soil water content estimation using electromagnetic waves with different frequency domains

To solve various problems that require geophysical countermeasures, several efforts have been made to estimate the soil water content through electromagnetic waves. This study was aimed at evaluating water content, and following three techniques of evaluation that use electromagnetic waves were compared based on the frequency domain: electrical resistivity measured at a low frequency, time-domain reflectometry (TDR) using a band of tens of MHz to several GHz, and a hyperspectral camera using light occupied in a very high frequency band. A multivariate regression model, which incorporated both the reflectance by hyperspectral images and dielectric constant as independent variables, demonstrated improved predictive accuracy compared with a single linear regression model, which solely focuses on reflectance. However, the relationship between the polynomials was induced by comparing wavelength bands that are superior to other methods when evaluating the water content of soil through hyperspectral technology. In particular, the reflectance at approximately 970 nm recorded the highest R-squared value when the curve fitted the cubic relationship between the reflectance and water content. As hyperspectral images offer the advantage of non-contact measurements, the soil water content was estimated using hyperspectral imaging techniques that exhibit similar high R-squared value as that of TDR.